The command center for AI-powered engineering
Fleet gives engineering teams a single binary to orchestrate hundreds of AI agents — with the governance, observability, and control that enterprise environments demand.
Our Mission
AI is transforming software development, but managing fleets of AI agents is still chaos — scattered terminals, no coordination, no visibility, no guardrails. Fleet is the command center that turns that chaos into a governed, observable, autonomous engineering operation.
Why Fleet Exists
The problem we set out to solve
Agent Sprawl
Teams spin up dozens of AI agents across terminals with no central visibility. Nobody knows what's running, what's stuck, or what's burning money.
Zero Coordination
Agents work in isolation. A developer agent creates a PR, but nothing tells a reviewer agent to look at it. Manual handoffs kill the speed advantage of AI.
No Guardrails
No budget limits, no approval gates, no audit trail. Engineering leaders can't adopt AI agents at scale without governance and observability.
Product Timeline
From first commit to enterprise-ready platform
First Commit
Core agent management — start, stop, and monitor AI agents as tmux sessions from a single terminal.
SQLite & Teams
Persistent state, team grouping, and the first CLI dashboards.
Public Launch
First public release. Binary downloads for macOS, Linux, and Windows.
Fabric Event Bus
Shared event bus for cross-agent coordination. Publish/subscribe across repos with claims, inbox, and task management.
MCP Integration
26 MCP tools over JSON-RPC 2.0. Claude Code, Cursor, and any MCP client can manage the fleet natively.
Pipeline Engine
Multi-stage pipeline orchestration with approval gates, rejection routing, and automatic retries.
Brain Daemon
Event-driven eval scoring, logistic regression risk model, anomaly detection, and auto-quarantine.
Enterprise
Audit logging, per-agent budget enforcement, org hierarchy, and air-gapped deployment support.
Fleet 1.13
136 agent templates, reactive event chains, watcher daemon, and the full org-centric architecture.
By the Numbers
What ships inside every Fleet release
What We Believe
The principles behind every decision
Simplicity
One binary. No Docker. No Node.js. No YAML orchestration layers. Complexity is a bug we refuse to ship.
Your Infrastructure
Self-hosted, air-gapped, zero telemetry. Your code never leaves your machines. No cloud dependency, ever.
Enterprise Governance
Budget controls, audit logs, approval gates, and org hierarchy. Fleet management that your security team will approve.
Developer Experience
If it takes more than 30 seconds to get running, we've failed. The CLI and MCP tools are designed for speed.
Get in Touch
We'd love to hear from you